“Loopers”在我脑海中挥之不去 @scaling01: Astra 6.2 的 NCRI / 串行推理深度比我预期的要高 所以如果它和 Astra 6 是同一个底层模型,那它可能更“loopy” 但我默认假设它们都基于同一模型,并回环两次 为什么? 对循环块再进行一次传递会增加成本,但价格从 Astra 6 到 6.2 下降了。而且 Astra 6.2 的 3 和 4 loop 配置将执行 > 240 层,这与 Jakub 的说法相矛盾。 现在,还有哪些选项是开放的? - Astra 6 训练不足,尚未有效利用 loopies(<-- 我最喜欢的) - Astra 6.2 是一个更大的模型 - 我搞错了,这一切都是错的
the loopers live rent free in my head
【引用 @scaling01】NCRI / serial reasoning depth for Astra 6.2 is a higher than I expected
so if it's the same underlying model as Astra 6 then it might be more loopy
but my default assumption is that they are both based on the same model and loop back two times
why?
another pass over the recurrent block would increase cost, but prices fell from Astra 6 to 6.2. And the 3 and 4 loop config for Astra 6.2 would execute > 240 layers, which would contradict Jakub's statement.
now, what options are still open?
- Astra 6 was undertrained and didn't yet utilize the loopies effectively (<-- my favorite)
- Astra 6.2 is a larger model
- I'm regarded and all of this is wrong
NCRI / serial reasoning depth for Astra 6.2 is a higher than I expected so if it's the same underlying model as Astra 6 then it might be more loopy but my default assumption is that they are both based on the same model and loop back two times why? another pass over the recurrent block would increase cost, but prices fell from Astra 6 to 6.2. And the 3 and 4 loop config for Astra 6.2 would execute > 240 layers, which would contradict Jakub's statement. now, what options are still open? - Astra 6 was undertrained and didn't yet utilize the loopies effectively (<-- my favorite) - Astra 6.2 is a larger model - I'm regarded and all of this is wrong在 X 查看被引用的帖子
来源:Lisan al Gaib · x.com